​
When to use sticky sessions

Sticky sessions keep the same residential exit IP for a chain of requests. Enable them when:
  • Your target site uses session cookies tied to IP (logged-in scraping)
  • The target has per-IP anti-bot scoring that punishes IP rotation (Cloudflare Bot Management, Akamai)
  • You’re doing multi-step checkout flows (add-to-cart → checkout → payment — same IP throughout)
  • You’re scraping a pagination chain and don’t want the site to see “different user” for each page
Don’t use them when:
  • The target site rate-limits per-IP — rotating IPs is the correct strategy (omit the session tag)
  • You’re scraping a static resource — no session state to preserve

​
Syntax

Append -session-<tag> to the username portion of your proxy auth: 
avp_live_<key>-session-<tag>:<secret>
Combine with country pinning: 
avp_live_<key>-country-us-session-<tag>:<secret>

​
Tag rules

RuleValue
Allowed chars[a-z0-9_-] case-insensitive
Max length64 chars
Scope(keyId, tag) pair — different keys with the same tag are separate sessions

​
Lifetime

The sticky binding lives 10 minutes from the last request that uses it. Every request with the same tag refreshes the 10-minute window — so a continuously-used session never expires. If the underlying exit node goes offline (WebSocket disconnects), the sticky binding is dropped and the next request picks a fresh node. Your session tag keeps working — it just binds to a different exit node silently.

​
Examples

​
One sticky session per scraper run

run_id = f"product-scrape-{date.today().isoformat()}"
proxy = f"http://avp_live_<key>-country-us-session-{run_id}:<secret>@api.atlasvpn.live:7777"

# All requests today use the same US exit IP
for sku in skus:
    requests.get(f"https://target.com/sku/{sku}", proxies={"https": proxy})

​
One sticky session per logged-in account

accounts = [('alice', 'pw1'), ('bob', 'pw2')]
for username, password in accounts:
    # Unique session per account — they each look like a different user
    auth = f"avp_live_<key>-country-us-session-acct-{username}:<secret>"
    proxy = f"http://{auth}@api.atlasvpn.live:7777"
    session = requests.Session()
    session.proxies = {"https": proxy}
    session.post("https://target.com/login", data={"u": username, "p": password})
    # Continue scraping as `alice` through a single consistent exit IP
    for url in my_scrape_plan:
        session.get(url)

​
Per-worker session in a concurrent scraper

import concurrent.futures
import uuid

def fetch(url):
    tag = uuid.uuid4().hex[:16]
    proxy = f"http://avp_live_<key>-session-{tag}:<secret>@api.atlasvpn.live:7777"
    return requests.get(url, proxies={"https": proxy})

with concurrent.futures.ThreadPoolExecutor(max_workers=20) as ex:
    results = list(ex.map(fetch, urls))
Each worker gets a fresh tag → fresh exit IP → parallelism without stepping on the rate limit.

​
Troubleshooting

SymptomCauseFix
Each request shows a different IP despite the tagTag contains invalid charsStick to [a-z0-9_-] — no spaces, no slashes
Sticky session “stops working” after ~10 minTTL expiredFire a keep-alive request every 5 min, or refresh on demand
502 after the session was workingNode went offlineRetry — the next request binds to a fresh node
Different sessions seem to share an IPYou ran out of nodes in that countryPool depth is beta-scale; happens in thin-pool countries